Method for protecting privacy under condition of MapReduce data processing frameworks

Abstract

The invention provides a method for protecting the privacy under the condition of MapReduce data processing frameworks, and belongs to the field of big data. An executing procedure of the method includes steps of (1), allowing users to define the maximum output ranges MaxRange of functions Mapper according to particularity of application environments; (2), judging whether results computed according to the maximum output ranges MaxRange and the functions Mapper are within the maximum output ranges MaxRange or not; (3), adding Laplacian noise to the output results of the functions Mapper according to differential privacy protection formulas if the results are within the maximum output ranges, or randomly selecting a number from the maximum output ranges MaxRange to be used as an output result of each function Mapper if the corresponding result is not within the corresponding maximum output range. The method has the advantages that the quantity of the functions Mapper for executing differential privacy protection can be reduced, and accordingly the algorithm running time can be shortened; the problem of excessively loud noise due to excessively high or low output valves of existing functions Mapper can be solved, requirements of differential privacy protection rules can be met, and the query precision can be improved.

Description

technical field [0001] The invention relates to the field of big data, in particular to a method for protecting privacy under the framework of MapReduce data processing. Using the method of adding Laplacian noise in the differential privacy protection strategy to meet the privacy protection of Mapper output results and using mandatory range checking to eliminate malicious code. Background technique [0002] Today, the development of social informatization and networking has led to explosive growth of data. According to statistics, only in terms of China's Internet data volume, Baidu visits more than 1 billion times a day, maintains and indexes more than 100 billion web pages, has more than 500 million active social users every day, and shares 4.5 billion pictures. The transaction volume has exceeded 20 billion, and the daily transaction peak has reached 100 million times. At the same time, various industries such as scientific computing, medical and health care, finance, a...

AI Technical Summary

Problems solved by technology

However, these methods are basically traditional data privacy protection technologies, which are suitable for small-scale, relational databases and file systems, and are not suitable for the current MapReduce computing framework.

Images